Schibsted – a perfect match for their thesis

Alexandra Hägg and Amanda Flöjs, master students in Industrial Statistics at Umeå Universitet.

How did two Industrial Statistics master students from Umeå University in Sweden experience writing their master thesis at Schibsted during the spring of 2020?

Are you a student with your master thesis ahead of you, or are you simply looking for an exciting opportunity to get professional development? Here’s your chance to be inspired by Alexandra Hägg and Amanda Flöjs, two master students in Industrial Statistics at Umeå University, who wrote their master thesis at Schibsted during the spring of 2020.

Despite the ongoing pandemic, not only did their project have the potential to create a lot of value for Schibsted and a Swedish newspaper within Schibsted, it also turned out to be a great personal experience and professional opportunity for the two students.

When Alexandra and Amanda initially started discussing what kind of thesis they wanted to write, they concluded to do a project within data (since their master focused on data analysis and statistical modelling) with the customer in focus. Schibsted, with its incredible amount of data and customers, was rapidly considered a perfect match.
They were given the chance of working with Schibsted’s Machine Learning team in collaboration with a Swedish newspaper within Schibsted. The topic of the thesis was to try to develop a statistical model to learn more about the newspaper’s digital subscription service churn. In other words, by using predictive models they tried to identify users’ behavior patterns that drive them to unsubscribe.

After several weeks in the Stockholm office and a couple of weeks of remote working (due to the pandemic), they came up with some interesting conclusions. With the help of statistical models their work showed that it is possible to predict which users are in the risk zone of ending their subscription. However, it is difficult for the model to generalize a specific behavioral pattern for the “active subscribers” and those who choose to unsubscribe.
“As long as the model is trying to predict human behavior, it will always be difficult,” Alexandra comments.

They also saw that the variables describing the frequency of use (i.e. how often a user uses the service) had more validity than variables describing the user’s activity in volume (i.e. how many articles a user reads). These are all very useful insights for the Swedish newspaper.

In addition to the results of the thesis results, the experience was fulfilling in several ways. We asked Alexandra and Amanda about their time in Schibsted:

What has it been like writing your master thesis at Schibsted?
Amanda: ”It has been extremely fun and developing. We felt welcome since day one, and we are so thankful to our supervisor who supported us all the way.”
Alexandra: ”I totally agree with Amanda. In addition, we worked with an exciting project that made the weeks go so fast.”

What did a “normal” working day look like for you?
Alexandra: The first six weeks we were in Schibsted’s Stockholm office, where we started each morning with a cup of coffee to go through the TODO’s. Then we often had a sync with either Patrik, our supervisor at the university or someone else at Schibsted. The rest of the days was all about modeling, analyzing and drinking loads of coffee. When we moved home due to the corona pandemic, the days looked very much the same. The only difference was that it took place in either Amanda’s or my apartment instead.

How has the corona pandemic affected your work?
Alexandra: After six weeks, the corona pandemic forced the whole Machine Learning team to work from home. It definitely affected us and our way of working together with them. However, working remotely and running meetings via video calls has gone surprisingly smooth.
Amanda: The result of our thesis was not that much affected by the pandemic. However, the pandemic has delayed the testing of our model in the business.

What was it like collaborating with your supervisor?
Amanda: Our awesome supervisor, Patrik Trelsmo is part of the Machine Learning team. We had regular meetings every week where we discussed and got help to sort out all our thoughts around the project and how to approach different problems. In addition to that, Patrik was always available for spontaneous questions or to support us in programming questions.
Alexandra: Yes, and I would also like to emphasize that everyone in the Machine Learning team has been very supportive and helpful.

Do you have any advice for students with a master’s thesis ahead of them?
Amanda: When it comes to choosing your project and company, my best advice is to choose the option that has a well-defined project and where you will have a committed supervisor who knows that he or she will have time for you.
Alexandra: I couldn’t agree more. I also advise you to dare to trust your university knowledge and not to be afraid to ask if there is something you do not understand!

What is your next adventure?
Alexandra: I will start as a Schibsted trainee in August and my first placement will be at Aftenposten in Oslo
Amanda: And I will also be a part of a Trainee program at SEB.

Thank you so much for the amazing work you have done. We wish you the best of luck!

Meet more of our people and read more Schibsted stories in Schibsted Future Report